readme

The subroutines glkern.f and lokern.f use an efficient and fast algorithm for automatically adapti

This is the README file for loading the F77-subroutines glkern.f and lokern.f
and/or the Splus function glkerns and lokerns and/or the matlab function
glkernm and lokernm (kernel regression estimation with automatically
adapted global bandwidth or local bandwidth array).

These algorithms were developped in the group of Theo Gasser by several
people, mainly Walter Koehler, Alois Kneip and Eva Herrmann

Version 10.1.1997

We are interested in any comments on the numerical and statistical
behaviour of these subroutines. Hence we will appreciate your comments.
Also any suggestion, how to improve the subroutines or how to implement
it into other software are welcome.

Questions and comments should be directed to

   Eva Herrmann
   Fachbereich Mathematik
   Technische Universitaet Darmstadt
   Schlossgartenstr. 7
   D-64289 Darmstadt

   Phone: ++49-6151-162494
   Fax:   ++49-6151-164011

   e-Mail: [e-mail]

 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
 !!!!! Please send an e-mail if you are interested in further updates !!!!
 !!!!! and informations on this and related routines.                 !!!!
 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Thank you!

Permission to use, copy, modify and distribute this software for any
non-commercial purpose is granted, provided the authors are appropriately
acknowledged.


 * The F77-subroutines glkern.f and lokern.f both need additional
   subroutines which are summarized in the file subs.f. They can be
   compiled and used directly as smoothing subroutines. They will write
   error messages and warnings on the standard output but this can be
   suppressed simply as is commented in the subroutine itself.
   A detailed description with general and computational remarks, statistical
   limits and examples can be found on the www-page
   http://www.unizh.ch/biostat/Software/kernf77.html.

 * For using the Splus functions glkerns and lokerns you have to follow
   these steps. After unpacking the kernS file you should have the following
   files besides this README file:

       glkerns.f                fortran code
       glkerns.s                interface function
       glkerns.d                Splus help file for glkerns
       glkerns.input            source file to create the functions
       lokerns.f                fortran code
       lokerns.s                interface function
       lokerns.d                Splus help file for lokerns
       lokerns.input            source file to create the functions

   To load the functions start Splus in this directory and run:

      > source("glkerns.input")
   and
      > source("lokerns.input")

   In every new Splus session you have to load the fortran routines again:

      > dyn.load("glkerns.o")
      > dyn.load("lokerns.o")

   A detailed description with general remarks, usage, examples and
   statistical remarks can be found on the www-page
   http://www.unizh.ch/biostat/Software/kernsplus.html

 * For using the matlab functions glkernm and lokernm you have to follow
   these steps. After unpacking the glkernm and lokernm file you should
   have the following files:

       glkernm.f                mex-interface and fortran code
       glkernm.m                M-file for help message
       lokernm.f                mex-interface and fortran code
       lokernm.m                M-file for help message

   Before starting matlab create a mex-file by calling

      fmex glkernm.f
   and
      fmex lokernm.f

   In every matlab session you can call glkernm and lokernm again.
   A detailed description with general remarks, usage, examples and
   statistical remarks can be found on the www-page
   http://www.unizh.ch/biostat/Software/kernmatlab.html